對自來水公司而言，自來水水質是銷售自來水所必需具備之品質保證書，亦是不可或缺的重要淨水操作參數，因此對其水質數據品質需嚴謹把關。除了需具備良好之品管措施，監控分析數據之準確性，更需藉由統計方法評估數據之合理性進一步確保數據品質。本研究即擬運用統計方法，評估數據合理性。以台灣自來水公司第六區管理處南化給水廠及烏山頭給水廠的水質監測數據進行統計分析，利用相關分析，找出相關性較高水質參數，再藉由迴歸分析，建立迴歸方程式，最後再進行驗證及評估。
本研究共建立30組迴歸方程式，南化給水廠14組，其決定係數（R 2）介於0.7222~0.9898，烏山頭給水廠16組，其決定係數（R 2）介於0.6410~0.9855。經96年數據驗證，部分水質參數預測值與實測值誤差可小於10%，故可有效評估數據合理性。惟當預測值與實測值有較大誤差時，需檢討檢測過程及進行數據查核，以驗證檢測數據合理性。另透過統計結果發現水質參數相關性不能一概而論，亦無法作一致性評估。建議台灣自來水公司應以統計方法，建立各淨水場水質數據評估機制，確保檢驗報告之檢驗數據正確性，為優質自來水品質保證。 For the Taiwan Water Corporation, water supply quality should be the warrantee for the water supplier and essential of important operation parameter for water purification. Including water quality control and the data reliability, furthermore, it should be to keep the data rationality by statistic methods. The research used statistic method to evaluate the rationality of water quality. Take the Nanhua Water Treatment Plant and Wushantou Water Treatment Plant of the Sixth branch of Taiwan as examples, to find the relationship among water quality parameters by correlation analysis, regression analysis and then evaluated them.
In this research, there were 30 sets of regression formula, including the 14 of those from Nanhua Water Treatment Plant and the others from Wushantou Water Treatment Plant. The R-squared values were 0.7222 ~ 0.9898 and 0.6410 ~ 0.9855 representatives. Part of the raw water quality is compared to predicted value of the regression formula, which is less than 10% error percentage. It evaluates effectively the rationality of water quality data but not for all water quality parameters. It should be review examination procedure and check the data. The correlation between water quality parameters isn’t always positive, so we recommend Taiwan Water Corporation should use statistic methods to set up data evaluation model for every purification plant, which make sure the accuracy of examination data report and guarantee water supply quality.
